|Publication number||US7805564 B2|
|Application number||US 10/898,259|
|Publication date||Sep 28, 2010|
|Filing date||Jul 26, 2004|
|Priority date||Dec 22, 1998|
|Also published as||US6542961, US6701410, US6701411, US6851029, US6910102, US7937527, US8051244, US8176248, US8375168, US20020091898, US20020095549, US20030191910, US20040158673, US20050033914, US20100005240, US20110167218, US20120011321, US20120203965|
|Publication number||10898259, 898259, US 7805564 B2, US 7805564B2, US-B2-7805564, US7805564 B2, US7805564B2|
|Inventors||Naoto Matsunami, Takashi Oeda, Akira Yamamoto, Yasuyuki Mimatsu, Masahiko Sato|
|Original Assignee||Hitachi, Ltd.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (61), Non-Patent Citations (2), Referenced by (24), Classifications (18), Legal Events (1)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This is a divisional application of U.S. Ser. No. 10/769,922, filed Feb. 3, 2004, now U.S. Pat. No. 6,910,102; which is a continuation application of U.S. Ser. No. 10/405,645, filed on Apr. 3, 2003, now U.S. Pat. No. 6,851,029; which is a continuation application of U.S. Ser. No. 10/095,581, filed Mar. 13, 2002, now U.S. Pat. No. 6,701,411; which is a continuation application of U.S. Ser. No. 09/468,327, filed on Dec. 21, 1999, now U.S. Pat. No. 6,542,961. This application is related to U.S. Ser. No. 10/095,578, filed Mar. 13, 2002.
This invention relates to a disk control system for controlling a plurality of disk devices and relates in particular to a method for improving the high speed operation of the disk control system, achieving a lower cost and improving the cost performance.
A diskarray system for controlling a plurality of disk devices is utilized as a storage system in computers. A diskarray system is for instance disclosed in “A Case for Redundant Arrays of Inexpensive Disks (RAID)”; In Proc. ACM SIGMOD, June 1988 (Issued by Cal. State Univ. Berkeley). This diskarray operates a plurality of disk systems in parallel and is a technique that achieves high speed operation compared to storage systems utilizing disks as single devices.
A method using the fabric of a fiber channel is a technique for mutually connecting a plurality of hosts with a plurality of diskarray systems. A computer system using this technique is disclosed for instance in “Serial SCSI Finally Arrives on the Market” of Nikkei Electronics, P. 79, Jul. 3, 1995 (No. 639) as shown in
In diskarray systems of the conventional art, when the number of disk devices were increased in order to increase the storage capacity and achieving a controller having high performance matching the number of disk units was attempted, the internal controller buses were found to have only limited performance and likewise, the processor performing transfer control was also found to have only limited performance. In order to deal with these problems, the internal buses were expanded and the number of processors was increased. However, attempting to solve the problem in this manner made the controller structure more complex due to the control required for a greater number of buses and caused increased overhead and complicated software control due to non-exclusive control of data shared between processors, etc. The rise in cost consequently became extremely high and performance reached its limits so that cost performance was unsatisfactory. Though the cost for this kind could be justified in terms of performance in a large scale system, in systems not on such a large scale the cost did not match performance, expandability was limited and the development period and development costs increased.
The overall system storage capacity and performance can be increased by connecting a plurality of diskarray systems in parallel with a fabric device. However, in this method, there is absolutely no connection between the diskarray systems, and access concentrated on a particular diskarray system cannot be distributed among the other devices so that high performance cannot be achieved to actual operation. Also, the capacity of a logical disk device (hereafter logic unit) as seen from the host is limited to the capacity of one diskarray system so that a high capacity logic unit cannot be achieved.
In an attempt to improve diskarray system reliability, a diskarray system can be comprised of a mirror structure where, in two diskarray systems, the host unit has a mirroring function. However, this method requires overhead due to control required of the mirroring by the host and also has the problem that performance is limited. This method also increases the load that the system administrator must supervise since many diskarray systems are present inside the system. The maintenance costs thus increase since a large number of maintenance personnel must be hired and maintenance fees must be paid for each unit. The plurality of diskarray systems and fabric devices are further all autonomous devices so that the settings must be made by different methods according to the respective device, creating the problem that operating costs increase along with a large increase in operating time and system administrator training time, etc.
In order to resolve these problems with the related art, this invention has the object of providing a disk storage system capable of being structured according to the scale and requirements of the computer system, and a disk storage system that responds easily to needs for high reliability and future expansion.
The disk storage system of this invention contains a storage device having a record medium for holding the data, a plurality of storage sub-systems having a controller for controlling the storage device, a first interface node coupled to a computer using the data stored in the plurality of storage sub-systems, a plurality of second interface nodes connected to any or one of the storage sub-systems, a switch connecting between a first interface node and a plurality of second interface nodes to perform frame transfer between a first interface node and a plurality of second interface nodes based on node address information added to the frame.
The first interface node preferably has a configuration table to store structural information for the memory storage system and a processing unit to analyze the applicable frame in response to the frame sent from the computer, converts information relating to the transfer destination of that frame based on structural information held in the configuration table, and transfers that frame to the switch. Further, when transmitting a frame, the first interface node adds the node address information about the node that must receive the frame, to that frame. A second interface node then removes the node address information from the frame that was received, recreates the frame and transfers that frame to the desired storage sub-system.
In the embodiment of this invention, the disk storage system has a managing processor connecting to the switch. The managing processor sets the structural information in the configuration table of each node according to the operator's instructions. Information for limiting access from the computer is contained in this structural information.
In another embodiment of this invention, the first interface node replies to the command frame sent from the computer instructing the writing of data, makes copies of that command frame and the following data frames, adds different nodes address information to each frame so the received frame and the copied command frames will be sent to the different respective nodes and sends these frames to the switch.
The hosts 30 and the diskarray subsets 10 are shown as four units each however this number is optional and is not limited. The hosts 30 and the diskarray subsets 10 may also be provided in different numbers of units. The diskarray switches 20 in this embodiment are duplexed as shown in the drawing. Each host 30 and each diskarray subset 10 are connected to both of the duplexed diskarray switches 20 by the respective host I/F31 and a diskarray I/F21. Thus even if one of the diskarray switches 20, the host I/F 31 or the diskarray I/F21 is broken, the other diskarray switches 20, the host I/F 31 or the diskarray I/F21 can be utilized to allow access from the host 30 to the diskarray system 1, and a high amount of usage can be achieved. However, this kind or duplication or duplexing is not always necessary and is selectable according to the level of reliability required by the system.
The host adapter 101 comprises an host MPU1010 to execute control of the adapter 101, an host system or in other words a diskarray I/F controller 1011 to control the diskarray switches 20 and the connecting I/F which is the diskarray I/F21, and an host bus 1012 to perform communications and data transfer between the cache memory/shared memory 102 and host MPU1010 and the diskarray I/F controller 1011. The figure shows one diskarray I/F controller 1011 for each host adapter 101 however a plurality of diskarray I/F controllers 1011 can also be provided for each one host adapter.
The lower adapter 103 contains a lower MPU103 to execute control of the lower adapter 103, a desk I/F controller 1031 to control the disk 104 and interface which is the disk I/F, and a lower bus 1032 to perform communications and data transfer between the cache memory/shared memory 102 and host MPU1030 and the diskarray I/P controller 1031. The figure shows four diskarray I/F controllers 1031 for each lower adapter 103 however the number of diskarray I/F controllers is optional and can be changed according to the diskarray configuration and the number of disks that are connected.
Each structural section of the diskarray switch 20 are preferably all comprised of hardware logic from the viewpoint of performance. However, program control utilizing general purpose processors is allowable for the SP2021 and the SC2022 functions if the specified performance can be achieved.
Each diskarray subset 10 has disk units 104 as one or a plurality of logical disk units. These logical disk units are referred to as Logical Units (LU). The LU need not correspond in a ratio of one to one, to the logical disk units 104 and one disk unit 104 can be comprised of a plurality of LU or one LU can comprise a plurality of disk units 104. One LU is recognized as one disk device as seen externally of the diskarray unit 10. In this embodiment, a logical LU is comprised further by a diskarray switch 20 and the host 30 functions to access this LU. In these specifications, when one LU is recognized as one LU by the host 30, then the LU is called independent LU (ILU) and when a plurality of LUs are recognized as one LU by the host 30, then the one LU recognized by the host 30 is called combined LU (CLU).
In this embodiment, when for instance the host #2 is accessing the region A1001, an access request is made specifying the region A1001, and this access request is converted by the diskarray switch 20 into a request for accessing the region A′ 1101 of the LU of the diskarray subset #0 and this request then sent to the diskarray subset #0. This diskarray subset #0 then performs access and mapping of the region A′ 1101 onto of the region A″ 1202 on the disk unit 104. The mapping between the address space 1000 and the address space 1100 is based on structural information held in the DCT207 in the diskarray switch 20. The details of this processing are related later on. The mapping performed in the diskarray subset is a technical method already well known in the prior art so a detailed explanation is omitted here.
In this embodiment, the DCT207 contains a Diskarray System Configuration Table and Diskarray Subset Configuration Tables. The structure of the Diskarray System Configuration Table is shown in
As shown in
The Host-LU Configuration Table 20271 has LU information (LU Info.) relating to the condition and Host LU of the diskarray subset 10 LU, which is information showing the LU type, CLU class, CLU stripe size and Host-LU indicating the affiliation of the LU and the Host-LU No. which is a number for identifying that LU. The LU Type in the table is information on the LU type showing that the Host-LU is a CLU or one LU. The CLU class is information showing the class is any one of “Joined”, “Mirrored” or “Striped” when the LU type of this Host-LU is shown to be a CLU. Here, “Joined” indicates as shown in
The diskarray I/F node configuration table 20272 contains information on what diskarray I/F node 202 of diskarray switch 20 is connected to each port of the diskarray subset 10 connected to the diskarray I/F 21. More specifically, this table holds the Subset NO. specifying the diskarray subset 10, the Subset Port No. specifying the port, the Switch No. specifying the diskarray switch 40 connected to that port, and an I/F Node No., specifying the diskarray I/F node 202 of the diskarray switch 20. When the diskarray subset 10 has a plurality of ports, information is set for each of those ports.
As shown in
The RAID Group Configuration Table 202730 has a Group No. showing the number added to the RAID Group, a level showing the RAID Level of that RAID Group, and Disks with information showing the number of disk comprising that RAID Group. When that RAID Group is comprised of striping such as for RAID Level 0, 5, then information showing that Stripe Size is included. As shown for instance, in the figure in the table, a RAID Group “0” is a RAID Group comprised of four disk units. The RAID Level is 5 and the Stripe Size is 0.
The LU Configuration Table 202740 has an LU No. showing the number (LUN) added to the LU, a RAID Group showing now that LU is configured in the RAID Group, a Condition showing the status of the LU, a Size showing the size (Capacity) of that LU, a Port showing what ports of the diskarray subset 10 are capable of providing access, and also an Alt. Port showing port that can be used as alternates for that Port No. The status showing the condition are of four types just as with the Host-LU and comprise “Normal”, “Warning” “Fault” and “Not Defined”. The port specified by information set in the Alt. Port is utilized when a problem occurs in the port specified with information set in the Port (item) however can also be used just for accessing the same LU from a plurality of ports.
The operation of the disk address system of this embodiment is described next.
In order to use the diskarray system, the setting of structural information of the diskarray subset 10 must be made for the diskarray switch 20. The system administrator can acquired structural setup information for the diskarray switch 20 and the diskarray subset 10 from a management console 5 by way of the diskarray configuration manager 70. The administrator can make different kinds of required entries of setup information such as logic unit structural setup for the desired system structure, RAID level settings, alternative path settings for use when trouble occurs. The diskarray configuration manager (means) 70 can receive that setting information, and transfer that setting information to the each diskarray subset 10 and diskarray switch 20. The entry of setup information on the management console 5 is described separately in the fifth embodiment.
In the diskarray switch 20, the communications controller 204 acquires the setup information and sets the structural information such as the address space information or each of the diskarray subsets 10 by means of the MP200. The MP200 distributes the structural information of the diskarray subset 10 to the each of the host I/F nodes 203 and the diskarray I/F nodes 202 by way of the crossbar switch 201. When the nodes 202 and 203 receive this information, the SP2021 stored this structural information in the DCT2027. In the diskarray subset 10, the diskarray subset configuration manager (means) 106 acquires the setup information and stores it in the shared memory 102. The host MPU1010 and the lower MPU 1030 refer to this setup information in the shared memory 102 and perform configuration management.
The operation when the read command is issued in described next for the diskarray system 1 with a host “#2”.
During reading of data, the host 30 issues a command frame “FCP_CMD” stored with the read command, to the diskarray switch 20 (arrow (a) in
When the CRC inspection shows that the information is correct, the SC2022 reads the frame held in the FB2021, recognizes this frame as the command frame, and analyzes the Frame Header 401 (step 20003). The SC2022 then instructs the SP2021 and registers the Exchange information such as S_ID, D_ID, OX_ID in the ET2026 (step 20004). Next, the SC2022 analyzes the frame payload 402 and acquires the LUN and CDB specified by the host 30 (step 20005). The SC2021 searches the DCT2020 at the instruction of the SC2022 and acquires the structural information of the diskarray subset 10. More specifically, the SC2021 searches the host-LU configuration table 20271 and finds information having a host-LU no. matching the LUN stored in the frame payload 402 that was received. The SC2021 recognizes the structure of the Host-LU from the information set in the LU Type, and CLU class, and based on the information held in the LU Info., identifies the disk subset 10 that must be accessed and its LUN in the LU as well as the LBA in the LU. Next, the SC2021 refers to the LU configuration table 202740 of the Diskarray Subset Configuration Table 202720 and confirms the connection port for the destination diskarray subset 10, and acquires from the Diskarray I/F Node Configuration Table 20272, the node No. of the diskarray I/F node 202 connected to that port. The SC2021 in this way acquires the convention information such as the No. LUN, LBA for recognizing the diskarray subset 10 and reports this information to the SC2022 (step 20006). Next, using the acquired conversion information, the SC2022 converts the LBA from the LUN and CDB of the frame payload 402. Also, the D_ID of the frame header 401 is converted to the D_ID of the host I/F controller 1011 of the diskarray subset 10. The S_ID is not rewritten at this point (step 20007). The SC2022 transfers the converted command frame and the diskarray I/F node 40, connected to the corresponding diskarray subset 10, to the SPG2024. The SPG2024 generates a packet added with a simple expansion header 601 such as shown in
The crossbar switch 201 receives the S Packet 60 from the SWP2010 connected to the host I/F node “#2”. The SWP2010 refers to the expansion header 601 of the S Packet 60, establishes a path for carrying out switch control for the SWP connecting with the transfer responder node, and transfers the S Packet 60 to the transfer responder of the diskarray I/F node 202 (Here, the diskarray I/F node “#0”). The SWP2010 establishes a path whenever the S Packet 60 is received and releases that path when transfer of the S Packet 60 is finished. In the diskarray I/F node “#0”, the SPG2024 receives the S Packet 60, removes the expansion header 601 and delivers the command frame portion to the SC2022. The SC2022 writes its own ID in the S_ID of the frame header of the command frame that was accepted. Next, the SC2022 instructs the SP2021 to register the Exchange information such as the S_ID, D_ID, OX_ID, of the command frame as well as the frame transfer originator host I/F node No. into the ET2026, and transfers this command frame to the IC2023. The IC2023 complies with instructions of the frame header 401 and transfers the command frame (arrow (b) of
The diskarray subset “#0” receives the command frame “FCP_CMD” after conversion, in the diskarray I/F controller 1011. The host MPU1010 acquires the LUN and CDB stored in the frame payload 402 of the command frame and recognizes that the LEN length data from the LBA of the specified logical unit is the read command. The host MPU1010 refers to the cache management information stored in the cache/shared memory 102 and performs cache miss-hit/ht identification. If a hit then the data is transferred from the cache 102. If a miss then reading of data from the disk unit is necessary so that address conversion is implemented based on the structure of RAID 5 and a cache space is secured. Processing information required for read processing from the disk unit 2 is generated, and processing information for continued processing in the lower MPU 1030 is stored in the cache/shared memory 102. The lower MPU 1030 starts processing when the processing information is stored in the cache/shared memory 102. The lower MPU 1030 specifies an appropriate disk I/F controller 1031 and generates a read command to the disk unit 2, and issued a command to the disk I/F controller 1031. The disk I/F controller 1031 stored the data read from the desk unit 2 in the address specified by the cache/shared memory 102 and issues a completion report to the lower MPU 1030. The lower MPU 1030 stores the processing completion report in the cache/shared memory 102 for reporting to the host MPU1010 that processing was completed correctly. The host MPU1010 restarts the processing when the processing completion report is stored in the cache/shared memory 102 and reports that read data setup is complete to the diskarray I/F controller 1011. The diskarray I/F controller 1011 issues a “FCP_XFER_RDY” which is a data transfer setup completion frame on the fibber channel for the applicable diskarray I/F node “#0” of the diskarray switch 20 (arrow (c) of
When the SPG2024 in the host I/F node “#2” receives the S Packet of the data transfer completion frame, the expansion header or the S Packet is removed, and the “FCP_XFER_RDY” reproduced and delivered to the SC2022 (step 20011). The SC2022 instructs the SC2021, searches the ET2026 and specifies the applicable Exchange (step 20012). Next, the SC2022 investigates whether the frame is “FCP_XFER_RDY” (step 20013) and if “FCP_XFER_RDY”, instructs the SP2021 to rewrite the originator Exchange ID (RX_ID) of ET2026. The value added to this frame is used as the originator Exchange ID (step 20014). The SC2022 then converts the S_ID, D_ID of the frame header 401 to an appropriate value used by the ID of the host 30 and the ID of the host I/F node 203 (step 20015) The frame header 401 is thus converted to a frame corresponding to the host “#2” by means of this processing. The IC2023 issues a “FCP_XFER_RDY” data transfer completion frame for this host “#2” (arrow (d) of
The diskarray I/F controller 1011 for the diskarray subset “#0” On generates a data frame “FCP_DATA” for performing data transfer, and transfers it to the diskarray switch 20 (arrow (e) of
Next, the diskarray subset “#0” of the diskarray I/F controller 1011 generates a status frame “FCP_RSP” to perform the end status transfer and issued this frame to the diskarray switch 20 (arrow (g) of
The read processing is thus performed from the diskarray. In the write processing for the diskarray system 1, only the transfer direction of the data frame is reverse and the processing is otherwise the same as the read processing.
The diskarray switch 20 as shown in
In the embodiment as described above, even if the performance of one diskarray subset unit is limited by the internal bus and the internal MPU, mutual connections can be made between the host and the diskarray subset by utilizing a plurality of the diskarray subsets, by means of the diskarray switch. In this way, high performance can be achieved as a total diskarray system. Even if the performance of a diskarray subset is relatively low, high performance can be attained by utilizing a plurality of diskarray subsets. Accordingly, low cost diskarray subsets can be connected in just the required number to match the scale of the computer system, and a diskarray system can be constructed at a cost appropriate to the desired scale. Further, when improvement in performance of increasing the disk capacity is required, then the diskarray subsets can be added in just the required amount. Still further, since a plurality of diskarray switches can be utilized to connect an optional number of hosts and diskarray subsets, a drastic improvement can be made in the capacity, the performance or the number of units for connection, and a system with high expandability obtained. Even still further, reduced elements of a diskarray system itself of the conventional art can be utilized in this embodiment so that large scale software that was previously developed can be utilized without changes, thus reducing development costs and achieving a short development period.
In the diskarray switches of this embodiment, it is assumed that a specified diskarray subset 10 can be accessed for each host I/F node 203. When one host I/F 31 is used in this case, only the LU110 in one diskarray subset 10 can be accessed. When accessing LU110 in a plurality of diskarray subsets 10 from one host unit is needed, then that host is connected to a plurality of host I/F nodes 203. Further, when setting access of LU110 of one diskarray subset 10 from a plurality of host 30, then loop topology or fabric topology can be utilized in the same host I/F node 203 to connect to the plurality of hosts 30. When configured in this way, during access of one LU110 from one host 10, a diskarray subset 10 can be set for each D_ID of the host I/F node 203 so that the LUN of each LU can be shown as is, to the host 30.
Since in this embodiment, the LU of each LU110 inside the diskarray subsets 10 can be shown unchanged to the host 30 for the above related reasons, then conversion of the LUN is no longer required in the diskarray switch 20. Accordingly, when the diskarray switch 20 receives a frame from the host 30, only the frame header 30 is converted the sa me as in the first embodiment, and the frame payload 402 is transferred without conversion to the diskarray subset 10. In the operation of each section of this embodiment, excluding the fact that the conversion of the frame payload 402 is not performed, the embodiment is the same as the first embodiment so that a detailed explanation of the identical sections is omitted. The diskarray switch 2 can be easily developed in this embodiment.
In the second embodiment, in the host I/F node of the diskarray switch, only the frame header 401 is converted, however in the third embodiment described hereafter, frame conversion, including the frame header is not performed. The computer system of this embodiment is configured the same as the computer system in the first embodiment as shown in
In the first and second embodiments, the internal structure of the diskarray system 1 such as the number of diskarray subsets 10 and the configuration of the LU110 are concealed from the host 30. The host 30 therefore sees the entire diskarray system 1 as one storage device. In contrast, in this embodiment, the diskarray subset 10 is revealed to the host 30, and the host 30 directly uses the D_ID of the Frame header as the port ID for the diskarray subset. By this arrangement the diskarray switch can control frame transfer just by complying with the frame header information, and the fabric of the fiber channel in the conventional art can be used instead of the diskarray switch 20 to achieve an equivalent switch device.
The diskarray system configuration manager (means) 70 communicates with the communication controller 106 of the diskarray subset 10 as well as the communication means 204 of the diskarray switch 20 and acquires or sets structural information of the diskarray subsets 10 and the diskarray switches 20.
The diskarray switches 20 have a structure basically the same as the diskarray switches of the first embodiment as shown in
In this embodiment, to achieve total management of the diskarray system structure with the diskarray system configuration manager 70, a diskarray management table (hereafter this table is called DCT, is provided in the diskarray system configuration manager 70. The DCT comprising the diskarray system configuration manager 70 consists of a group of two tables; a Diskarray System Configuration Table 20270 and a Diskarray Subset Configuration Table 202720-202723. The host-LU in this embodiment are all comprise as one LU so that the “LU Type” in the Host-LU-Configuration table 20271 are all “ILU”, and the “CLU Class” and CLU Stripe Size” are not significant.
The administrator operates the management console 5, communicates with the diskarray system configuration manager 70 and acquires information such as the number of disk units, and disk capacity of the diskarray subset 10, and performs setting of the LU110 of the diskarray subset 10 and setting of the RAID level. Next, the administrator communicates with the diskarray system configuration manager 70 from the management console 5, controls the diskarray switch 20 and sets related information among the host 30 and the diskarray subsets 10. This operation establishes the structure of the diskarray system 1 and allows LU1 to be seen as the administrator wishes, from the host 30. The diskarray system configuration manager 70 saves the above setting information, verifies the configuration according operation by the administrator and performs changes in the structure (configuration).
In this embodiment, once the diskarray system 1 is configured, a plurality of diskarray systems 1 can be handled the same as one diskarray system and without making the administrator aware of the presence of the diskarray switch 20. Further in this embodiment, the diskarray subsets 10 and the diskarray switches 20 can be operated together by means of the same operating environment and confirming their configuration (or structure) and making changes in the configuration is also simple. Still further in this embodiment, when substituting the diskarray system of this embodiment with a diskarray system used in the conventional art, no changes are made in the host 30 settings, and the structure of the diskarray system 1 wean work with the diskarray system structure used up until then, and interchangeability can be maintained.
A fiber channel was used in the host I/F in the first through third embodiments described above. In the embodiment hereafter described, an interface other than the fiber channel might also be used.
The host 30 in this embodiment, issues a SCSI command to the diskarray I/F node 203. In the case of a read command, the SPC20230 stores this in the BUF 20232 and reports reception of the command by breaking into the PEP 20231. The PEP 20231 uses the command stored in the BUF20232, and converts the command to FPC20233 and sends it to the FPC20233. When the FPC20233 receives this command, in convert the command into a frame configuration and delivers it to the SC2022. At this time, the Exchange ID, Sequence ID, Source ID and Destination ID are added to PEP 20231 capable of the following processing. The remaining command processing is performed the same as in the first embodiment When the setup of data is complete, the data array subset 10 issues a data transfer setup completion frame, and after the data transfer ends correctly, implements issue of a status frame. In the period from the diskarray subset 10 to the IC2023, while the frame header 401 and the frame payload 402 are being converted as required, the transfer of each frame is performed. The FPC20233 of the IC2023 receives the data transfer setup completion frame, then receives the data and stores it in the BUF 20232 and if the transfer has ended correctly, receives the status report, and breaks into the PTP20231 to report that transfer of data is complete. When the PTP20231 receives the break-in (interruption), the SPC20230 starts up and instructs the start of data transfer to the host 30. The SPC20230 transmits the data to the host 30, and after confirming normal completion, interrupts the PTP20231 to report the data transfer ended correctly.
A parallel SCSI was used as an example here of a host I/F other than a fiber channel however other interfaces wan be implemented such as for ESCON in the same manner as a host I/F to the main frame. Host I/F nodes corresponding for instance, to the fiber channel, parallel SCSI and ESCON can be provided as the host I/F node 203 of the diskarray switch 20 so that all kinds of so-called open systems such a personal computers and work stations can be connected with the main frame to one diskarray system 1. In this embodiment, a fiber channel was utilized as the diskarray I/F in the first through the third embodiments however the desired optional I/F can also be used as the diskarray I/F.
A method for configuration management of the diskarray system 1 is described using the fifth embodiment.
Next, the addition of another diskarray subset 10 is described. When adding a new diskarray subset 10, the administrator connects the diskarray subset 10 to be added, to an empty I/F node 202 of the diskarray switch 20. The administrator next operates the management console 5 and presses the “Show Latest Status” button 5001 displayed on the logical connection configuration screen 50. A picture showing the diskarray subsets not yet set appears on the screen (not shown in drawing) in response to pressing the button 5001. When the picture for this diskarray subset is selected, the setup screen for the diskarray subsets then appears. The on this setup screen, the administrator executes the various settings for the newly added diskarray subset items set on this screen include the RAID level and the LU configuration. Next, on switching to the logical connection configuration screen of
Next the mirroring process by means of the diskarray switch 20 is described utilizing the sixth embodiment. The mirroring described here, is a method to support duplexed (duplicated) writing by means of two independent LU of two diskarray subsets, and duplicating including up to the controller of the diskarray subset. The reliability therefore is different from the method duplexing only the disks.
The system configuration (structure) of this embodiment is the same as shown in
The operation of each section of this embodiment is largely the same as the first embodiment. Hereafter, the points differing from the first embodiment are explained mainly with the operation of the host I/F node of the diskarray switch 20.
In the write operation, the write command frame (FCP_CMD) issued by the host 30 is received by the IC2023 (arrow (a) of
The diskarray subsets “#0” and “#1” receive the command frames and the respective, independent, data transfer setup completion frames “FCP_XFER_RDY” are distributed to the diskarray switch 20″ (arrows (c0) (c1) of
After receiving the data transfer setup completion frame, the host 30 sends the data frame (FCP_DATA) to the diskarray switch 20 (arrow (e) of
The diskarray subsets “#0” and “#1” receive each of the data frames, respectively write these frames in the disk unit 104, and set the status frame (FCP_RSP) to the diskarray switch 20. When the SP2022 receives the status frames from the respective diskarray subsets “#0” and “#1”, their respective expansion headers are removed from their status frames, the frame header restored and the exchange information acquired from the ET2026 (step 21041, 21042). When the status frames from both the diskarray subsets “#0” and “#1” are arranged (step 21043), conversion of the master status frame from the LU “#0” is performed (step 21044) after checking that the status has completed correctly, and the slave status frame is deleted (step 21045). Then, the IC2023 sends a command frame to the host to report correct completion (arrow (h) of
The write processing in the mirrored structure is thus completed. The read processing for the mirrored LU “#01” differs only in the direction of data transfer, and is performed largely the same as the above described write processing except that the issue of a read command to two diskarray subsets is not necessary, and a command frame can be issued just to either diskarray subset. A command frame for instance can be issued mainly to the master LU however for high speed operation, methods such as alternate issue of command frames for both the master/slave LU will prove effective in distributing the load.
In the above related processing, in steps 21014 and step 21043, a reply from the two diskarray subsets LU “#0” and “#1” is awaited, both synchronized with and the process then proceeds. With this kind of control, handling of errors is simple since the process proceeds after verifying the success of the processing for both of the diskarray subsets. On the other hand this kind of control has the drawback performance declines since the overall processing speed depends on which of the replies is slower. To resolve this problem, in the diskarray switch, control such as by proceeding to the next process without waiting for a reply from the diskarray subset or a “Asynchronous type” control that proceeds to the next process at the print where a reply from either one of the diskarray subsets is received are possible. The frame sequence when this asynchronous type control is used is shown by the dashed arrow lines in
The following processing is implemented when an error occurs during frame transfer between the diskarray subset 10 and the diskarray switch 20. When the process being implemented is write processing, then a retry process is performed on the LU in which the error occurred. If the retry process is a success, then the process continues unchanged. However, when the retry process fails after a preset number of retries, then the diskarray switch 20 prohibits access to this diskarray set 10 (or LU) and information showing this prohibition is registered in the DCT2027. The diskarray switch 20 also reports this information to the diskarray system configuration manager 70 by way of the communication controller 204 and the MP200. The diskarray system configuration manager 70 then issues an alarm to the management console 5 in response to this report. The administrator can thus recognize that trouble has occurred. Afterwards, the diskarray switch 20 continues the operation by utilizing a normal diskarray subset. The host 30 also continues processing without recognizing that an error has occurred.
This embodiment utilizes a mirror configuration in a two unit diskarray subsystem to that the disk is made more resistant to problems that occur. The resistance of the diskarray controller, diskarray I/F, and the diskarray I/F node can also be improved, and the reliability of the overall diskarray system can be improved without taking measures such as duplexing (duplicating) the internal buses.
In the seventh embodiment, a method is described for combining three or more diskarray subsets 10 and configuring them into one logical diskarray subset group. In this embodiment, data is distributed and stored into a plurality of diskarray subsets 10. Distributing and storing the data in this way allows distributing the access to the diskarray subsets, to prevent the access being concentrated in a particular diskarray subset so that the throughput of the total group is improved. A diskarray switch is used in this embodiment to implement this kind of striping.
An address map of the disk address system 1 of this embodiment is shown in
Hereafter, the operation of the host I/F node 203 in this embodiment is described while referring to the operation flowchart shown in
When a command frame is issued from the host 30, the diskarray switch 20 receives this command frame with the IC2023 of the host I/F node 203 (step 22001). The SC2022 accepts this command frame from the IC2023, searches the DCT2027 using the SP2021 and verifies that striping is necessary (step 22005). Next, SC2022 searches the DCT2027 using the SP2021, finds from the structural information containing the stripe size S, the stripe No. for the stripe belonging to the data being accessed, and designates what diskarray subset 10 this stripe is stored in (step 22006). Stripe crossover may possible occur at this time however this processing in such a case is related later. When no stripe crossover occurs, the SC2022 implements conversion of the command frame (step 22007) based on SP2020 calculation results, and stores the exchange information in the ET2026 (step 22008). The subsequent processing is the same as for the first embodiment.
When stripe crossover has occurred, the SP2021 generates two command frames. These frames are generated for instance, by duplicating the command frame issued from the host 30. New settings are made such as for the frame header and frame payload of the generated command frame. After duplicating the command frame in SC2022, conversion can also be implemented the same as in the sixth embodiment however in this embodiment is newly made by SP2022. Wlen the two command frames are made, the SC2022 sends there frames to the respective diskarray subsets 10. Data transfer is then performed the same as in the first embodiment. The point in this embodiment differing from the first embodiment is that the data itself must be transferred between one host 30 and two diskarray subsets 10. In the read process for instance, the data frame transferred from the two diskarray subsets 10, must be transferred to all the hosts 30. The SC2022 at this time, complies with the information registered in the ET2026, and adds the appropriate exchange information, in the appropriate order to the data frame transferred from the diskarray subset 10 and sends this to the host 30. In the write process, two data frames are made, the same as for the command frame, and transferred to the applicable diskarray subset 10. The sequential control of the data frames at the host or the diskarray subset is called the “Out of Order” function. This “Out of Order” function is not required if the configuration is compatible with nonsequential processing. Finally, when all data transfer is complete, and the diskarray switch 20 has received the status frames respectively from the two diskarray subsets 10, the SP2021 (or the SC2022) makes a status frame for the host 30, and the IC2023 sends this status frame to the host 30.
This embodiment as described above, is capable of distributing the access (load) into a plurality of diskarray subsets, so that along with improving the total throughput, the access latency can be reduced.
Next, the duplicating operation between the two diskarray systems (or the diskarray subsets) is described using the eighth embodiment. In the system described here, one of two diskarray systems is installed at a remote location to provide recovery assistance in case of damage to the other diskarray system due to a natural or man-made calamity, etc. This kind of countermeasure for dealing with damage from disasters is referred to as disaster recovery and the making of copies performed with the diskarray system at the remote location is referred to as remote copy.
In the mirroring as described in the sixth embodiment, the mirror function is achieved with the diskarray subsets 10 installed at largely the same location geographically so that diskarray I/F21 can use a fiber channel. However when diskarrays (diskarray subsets) are performing remote copy at remote locations in excess of 10 kilometers, then a fiber channel cannot be used to transfer a frame unless relay equipment is added. A mutual distance of some several hundred kilometers is used during disaster recovery so that use of fiber channels for connecting between diskarrays is impractical. Therefore methods such as satellite communications or high speed public telephone lines with ATM (Asynchronous Transfer Mode) are utilized.
The remote copy process in this embodiment is similar to the mirroring process in the sixth embodiment. The points differing from the mirroring process of the sixth embodiment are explained next. When the host 30 issues a write command frame, the diskarray system 10 of site A81 performs frame duplicating the same as in the sixth embodiment, and transfers one of the copied (duplexed) frames to its own diskarray subset 10. The other frame is converted from a fiber channel frame to an ATM packet by the ATM node 205 and sent to the site B82 by way of the public telephone line 9. At the site B82, the ATM node 205 of the diskarray switch 20 receives this packet. The IC2023 of the ATM node 205, restores the fiber channel frame from the ATM packet, and transfers the fiber channel frame to the SC2022. The SC2022 implements frame conversion the same as when the write command was received from the host 30 and transfers the frame to the remote copy diskarray subset. From hereon, fiber channel—ATM conversion is performed for all the data transfer setup completion frames, data frames and status frame, and by implementing the same frame transfer process, remote copy can be achieved. When the read command frame was issued from the host 30, the diskarray switch 20 transfers the command frame only to the diskarray subset 10 only for its own site and reads this data only from the diskarray subset 10 of its own site. The operation at this time is the same as in the first embodiment.
This embodiment is capable of making backups of user data in real-time and providing recovery assistance when damage has occurred to a diskarray system site due to a disaster, etc.
The combining of a plurality of LU in one diskarray subset 10 is described next. The disk storage device for a main frame for instance, has a logical volume size set to a maximum value of 2 GB in order to maintain interchangeability with the previous system. When using this kind of diskarray system as an open system, the LU receive the same restrictions on the logical volume size, so that the hosts see this configuration as a large number of small size LU. This kind of method has the problem that operating the system is difficult when the system has developed to a high capacity level. To deal with this problem, a method was contrived for combining these logical volume (in other words LU) units into one large combine LU (CLU) structure by means of the diskarray switch 20. The forming of a combined LU (CLU) is achieved in this embodiment by the diskarray switch 20. The combining of LU in this embodiment is the same as the forming of combined LU by means of a plurality of diskarray subsets 10 in the first embodiment. The differing point is only that in this embodiment, a plurality of LU are combined within the same diskarray subset 10. The operation as a diskarray system is completely the same as in the first embodiment.
By combining a plurality of LU in the same diskarray subset 10 in this way, to form one large LU, a diskarray system is achieved having excellent operability, reduced management cost and in which there is no need for the host to manage a large number of LU.
Next, a method for setting alternative paths by means of the diskarray switch 10 is explained while referring to FIG 24. The structure of each section in the computer system shown in
Here, the alternative path for the diskarray I/F “#0”, is set as the diskarray I/F “#1”, while the alternative path for the diskarray I/F “#1” is set as the diskarray I/F “#0”1. Alternative paths are set in the same way respectively for the host adapter in the diskarray subset 10, the cache memory/shared memory, and the lower adapter.
Next, the setting of the alternative path is described, assuming that a problem has occurred and the path connecting the diskarray I/F21 to the host adapter “#1” of the diskarray subset 1 is broken or unusable as shown in
The embodiment described above, can therefore switch to an alternative path when a problem occurs on a path, without this switch being recognized by the host and render the setting of substitutes on the host side unnecessary. Thus the utilization of the system can be improved.
In this invention as described above, a storage system can be achieved that easily improves the storage device expandability, and reliability according to various requirements and the scale of the computer system. The above explanations of the each of the embodiments all utilized a diskarray system having a disk device. However, this, invention is not limited to use of a disk device as a storage media and is also applicable to optical disk devices, tape devices, DVD devices and semiconductor storage devices, etc.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5140592||Oct 22, 1990||Aug 18, 1992||Sf2 Corporation||Disk array system|
|US5155835 *||Nov 19, 1990||Oct 13, 1992||Storage Technology Corporation||Multilevel, hierarchical, dynamically mapped data storage subsystem|
|US5193184 *||Jun 18, 1990||Mar 9, 1993||Storage Technology Corporation||Deleted data file space release system for a dynamically mapped virtual data storage subsystem|
|US5257391||Aug 16, 1991||Oct 26, 1993||Ncr Corporation||Disk controller having host interface and bus switches for selecting buffer and drive busses respectively based on configuration control signals|
|US5293635 *||Apr 30, 1991||Mar 8, 1994||Hewlett-Packard Company||Detection on a network by a mapping application of a relative location of a first device to a second device|
|US5390173 *||Oct 22, 1992||Feb 14, 1995||Digital Equipment Corporation||Packet format in hub for packet data communications system|
|US5430855 *||Aug 31, 1994||Jul 4, 1995||Storage Technology Corporation||Disk drive array memory system using nonuniform disk drives|
|US5509018 *||Sep 10, 1993||Apr 16, 1996||International Business Machines Corporation||Flash-erase-type nonvolatile semiconductor storage device|
|US5517662||Nov 8, 1994||May 14, 1996||International Business Machines Corporation||Multiprocessor system with distributed memory|
|US5519844||Mar 21, 1994||May 21, 1996||Emc Corporation||Logical partitioning of a redundant array storage system|
|US5559958 *||Aug 20, 1993||Sep 24, 1996||Compaq Computer Corporation||Graphical user interface for computer management system and an associated management information base|
|US5568629 *||Nov 4, 1993||Oct 22, 1996||At&T Global Information Solutions Company||Method for partitioning disk drives within a physical disk array and selectively assigning disk drive partitions into a logical disk array|
|US5581735||Feb 17, 1994||Dec 3, 1996||International Business Machines Corporation||System for supplying unit image data to requesting users from multiple storage devices based on directory information and token queues identifying the requester and data|
|US5606359||Jun 30, 1994||Feb 25, 1997||Hewlett-Packard Company||Video on demand system with multiple data sources configured to provide vcr-like services|
|US5619690||Jun 21, 1994||Apr 8, 1997||Hitachi, Ltd.||Computer system including a computer which requests an access to a logical address in a secondary storage system with specification of a local address in the secondary storage system|
|US5729763||Aug 15, 1995||Mar 17, 1998||Emc Corporation||Data storage system|
|US5745789||Jan 19, 1993||Apr 28, 1998||Hitachi, Ltd.||Disc system for holding data in a form of a plurality of data blocks dispersed in a plurality of disc units connected by a common data bus|
|US5751965 *||Mar 21, 1996||May 12, 1998||Cabletron System, Inc.||Network connection status monitor and display|
|US5754791 *||Mar 25, 1996||May 19, 1998||I-Cube, Inc.||Hierarchical address translation system for a network switch|
|US5768520 *||Jan 29, 1996||Jun 16, 1998||International Business Machines Corporation||Method for determining load capacity by grouping physical components into logical components whose loads represent fixed proportional loads of physical components|
|US5812754 *||Sep 18, 1996||Sep 22, 1998||Silicon Graphics, Inc.||Raid system with fibre channel arbitrated loop|
|US5835694||Dec 6, 1996||Nov 10, 1998||International Business Machines Corporation||Raid-configured disk drive array wherein array control occurs at the disk drive level|
|US5974237 *||Dec 18, 1996||Oct 26, 1999||Northern Telecom Limited||Communications network monitoring|
|US5999179 *||Nov 17, 1997||Dec 7, 1999||Fujitsu Limited||Platform independent computer network management client|
|US6009466 *||Oct 31, 1997||Dec 28, 1999||International Business Machines Corporation||Network management system for enabling a user to configure a network of storage devices via a graphical user interface|
|US6023780||May 6, 1997||Feb 8, 2000||Fujitsu Limited||Disc array apparatus checking and restructuring data read from attached disc drives|
|US6078990||Feb 6, 1998||Jun 20, 2000||Ncr Corporation||Volume set configuration using a single operational view|
|US6098119||Jan 21, 1998||Aug 1, 2000||Mylex Corporation||Apparatus and method that automatically scans for and configures previously non-configured disk drives in accordance with a particular raid level based on the needed raid level|
|US6105122||Feb 6, 1998||Aug 15, 2000||Ncr Corporation||I/O protocol for highly configurable multi-node processing system|
|US6128016 *||Dec 20, 1996||Oct 3, 2000||Nec Corporation||Graphic user interface for managing a server system|
|US6148349||Feb 6, 1998||Nov 14, 2000||Ncr Corporation||Dynamic and consistent naming of fabric attached storage by a file system on a compute node storing information mapping API system I/O calls for data objects with a globally unique identification|
|US6173374||Feb 11, 1998||Jan 9, 2001||Lsi Logic Corporation||System and method for peer-to-peer accelerated I/O shipping between host bus adapters in clustered computer network|
|US6209033 *||May 13, 1997||Mar 27, 2001||Cabletron Systems, Inc.||Apparatus and method for network capacity evaluation and planning|
|US6247077||Feb 6, 1998||Jun 12, 2001||Ncr Corporation||Highly-scalable parallel processing computer system architecture|
|US6256740||Feb 6, 1998||Jul 3, 2001||Ncr Corporation||Name service for multinode system segmented into I/O and compute nodes, generating guid at I/O node and exporting guid to compute nodes via interconnect fabric|
|US6263374||Aug 4, 1994||Jul 17, 2001||International Business Machines Corporation||Apparatus for coupling a bus-based architecture to a switch network|
|US6289375 *||Oct 30, 1998||Sep 11, 2001||International Business Machines Corporation||Method and apparatus for invoking network agent functions using a hash table|
|US6314460 *||Oct 30, 1998||Nov 6, 2001||International Business Machines Corporation||Method and apparatus for analyzing a storage network based on incomplete information from multiple respective controllers|
|US6329985 *||Jun 30, 1998||Dec 11, 2001||Emc Corporation||Method and apparatus for graphically displaying mapping of a logical object|
|US6421711 *||Jun 29, 1998||Jul 16, 2002||Emc Corporation||Virtual ports for data transferring of a data storage system|
|US6466973 *||Mar 6, 1998||Oct 15, 2002||Adaptec, Inc.||Method and system for managing storage devices over a network|
|US6484245||May 28, 1998||Nov 19, 2002||Hitachi, Ltd.||Apparatus for and method of accessing a storage region across a network|
|US6493750||Oct 30, 1998||Dec 10, 2002||Agilent Technologies, Inc.||Command forwarding: a method for optimizing I/O latency and throughput in fibre channel client/server/target mass storage architectures|
|US6542961||Dec 21, 1999||Apr 1, 2003||Hitachi, Ltd.||Disk storage system including a switch|
|US6664978 *||Nov 17, 1997||Dec 16, 2003||Fujitsu Limited||Client-server computer network management architecture|
|US6839747 *||Jun 30, 1999||Jan 4, 2005||Emc Corporation||User interface for managing storage in a storage system coupled to a network|
|US6910102 *||Feb 3, 2004||Jun 21, 2005||Hitachi, Ltd.||Disk storage system including a switch|
|US7107534 *||Mar 24, 1999||Sep 12, 2006||Adaptec, Inc.||Storage area network administration|
|US20020002606 *||Mar 6, 1998||Jan 3, 2002||David H. Jaffe||Method and system for managing storage devices over a network|
|US20050251620 *||May 2, 2005||Nov 10, 2005||Hitachi, Ltd.||Data migration in storage system|
|JPH0713917A||Title not available|
|JPH05197498A||Title not available|
|JPH05265661A||Title not available|
|JPH06187201A||Title not available|
|JPH08171463A||Title not available|
|JPH08328760A||Title not available|
|JPH09198308A||Title not available|
|JPH09292954A||Title not available|
|JPH09305328A||Title not available|
|JPH10333839A||Title not available|
|JPS648433U||Title not available|
|1||Microsoft Computer Dictionary, Forth Edition, definition of the term "node", p. 312, 1999.|
|2||Serial SCSI Finally Arrives on the Market, Nikkei Electronics, No. 639, 1995,p. 79.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US8510519 *||Aug 2, 2010||Aug 13, 2013||Taejin Info Tech Co., Ltd.||Hybrid raid controller|
|US8510520 *||Aug 2, 2010||Aug 13, 2013||Taejin Info Tech Co., Ltd.||Raid controller having multi PCI bus switching|
|US8645662 *||Jan 25, 2012||Feb 4, 2014||Oracle International Corporation||Sub-lun auto-tiering|
|US8819116 *||Mar 8, 2010||Aug 26, 2014||Amazon Technologies, Inc.||Providing services using a device capabilities service|
|US8819316 *||Jun 21, 2011||Aug 26, 2014||Taejin Info Tech Co., Ltd.||Two-way raid controller with programmable host interface for a semiconductor storage device|
|US8856258 *||Mar 1, 2010||Oct 7, 2014||Samsung Electronics Co., Ltd.||Host device and slave device controlling method|
|US8887655||Jan 25, 2012||Nov 18, 2014||Honeywell International Inc.||Valve actuator with position indicator extension|
|US9032993||Jun 14, 2012||May 19, 2015||Honeywell International Inc.||Handle mechanism for an HVAC damper actuator|
|US9176670 *||Apr 26, 2011||Nov 3, 2015||Taejin Info Tech Co., Ltd.||System architecture based on asymmetric raid storage|
|US9423143||Dec 18, 2013||Aug 23, 2016||Honeywell International Inc.||HVAC actuator with light indicator|
|US9460030||Sep 5, 2014||Oct 4, 2016||Samsung Electronics Co., Ltd.||Host device and slave device controlling method|
|US9485146||Jul 15, 2014||Nov 1, 2016||Amazon Technologies, Inc.||Providing services using a device capabilities service|
|US9568207||Dec 18, 2013||Feb 14, 2017||Honeywell International Inc.||HVAC actuator with removable wire blocking tab|
|US9623523||Dec 18, 2013||Apr 18, 2017||Honeywell International Inc.||HVAC actuator with taping flange|
|US9664409||Jun 14, 2012||May 30, 2017||Honeywell International Inc.||HVAC damper system|
|US20110047239 *||Mar 1, 2010||Feb 24, 2011||Samsung Electronics Co., Ltd.||Host device and slave device controlling method|
|US20120030410 *||Aug 2, 2010||Feb 2, 2012||Byungcheol Cho||Hybrid raid controller|
|US20120030417 *||Aug 2, 2010||Feb 2, 2012||Byungcheol Cho||Raid controller having multi pci bus switching|
|US20120278526 *||Apr 26, 2011||Nov 1, 2012||Byungcheol Cho||System architecture based on asymmetric raid storage|
|US20120278527 *||Apr 26, 2011||Nov 1, 2012||Byungcheol Cho||System architecture based on hybrid raid storage|
|US20120278550 *||Apr 26, 2011||Nov 1, 2012||Byungcheol Cho||System architecture based on raid controller collaboration|
|US20120331338 *||Jun 21, 2011||Dec 27, 2012||Byungcheol Cho||Two-way raid controller with programmable host interface for a semiconductor storage device|
|US20130166839 *||Jan 25, 2012||Jun 27, 2013||Oracle International Corporation||Sub-lun auto-tiering|
|USD728071||Dec 27, 2013||Apr 28, 2015||Honeywell International Inc.||HVAC actuator|
|U.S. Classification||711/114, 709/224, 711/206, 370/244, 711/156, 715/736, 711/202|
|International Classification||G06F13/00, G06F13/20, G06F12/00|
|Cooperative Classification||G06F3/0611, G06F3/0635, G06F3/0689, G06F3/067|
|European Classification||G06F3/06A4C6, G06F3/06A6D, G06F3/06A2P2, G06F3/06A6L4R|